Masking Images
Segmentation, Thresholding and Creating Binary Masks
Although Image Masks can be created by manually tracing around a cell's edge, this
approach is labor intensive for a single image let alone many multi-plane image stacks.
Therefore for image processing, masking is usually a semi-automated endeavor.
Images are first Segmented on the basis of pixel Intensity; this must be done on a
plane-by-plane basis for an image stack. In Metamorph the segmented overlay planes
are then extracted to a separate image stack for the purpose of creating a Binary Mask.
A Binary mask exclusively has pixel intensity values of “0” (background) or “1”
(object). Multiplying the I-SO and EGFP Images with their conjugate Binary
Mask results in masked-images which only contain values (above zero) for cellular
regions.
Masking images allows the Image Processor to discriminate Regions of Interest; these
may also be referred to as Areas of Interest or Objects. A second reason to mask
images is to prevent artifacts arising from dividing background by background values.
Considerations:
•
•
•
When Thresholding an Image Stack remember that cells were brighter at earlier
time points; inspect the threshold max values for the entire data set.
Set minimum intensity threshold values (for Segmentation) that do not include
much (or any) background regions; this is sometimes difficult.
Make sure that the resulting mask is a stack of images with the same number of
planes (time points) as the image sets.
Left:
An image to be thresholded as a function of pixel intensity
Right: The image has been Segmented and the pink overlay corresponds to
the location of the Cell.
Bottom: The Thresholding dialog box (Metamorph); a range of intensity
values between 574 and 1597 were selected to Segment the Cell.
Left:
A 1-Bit Mask (having pixel intensity values of 1 (White) and 0 (Black)
was created from Overlay Image (Pink) above.
Right: The Image to be masked is pseudo-colored; the background values
are very low but are above zero.
Bottom: The Masked Image is the product of the 1-Bit Binary Mask (Left) and
the Image (Right).